Wholesale and retail distribution companies that have implemented demand forecasting solutions aren’t always getting the most out of their investment.
A recent article on the DC Velocity website highlights why most companies aren’t reaping the full benefits of demand forecasting software. Shaun Snapp, an expert in the field, says demand forecasting is a decades-old concept that really became popular in 2006 as computers became advanced enough to make the calculations needed for advanced supply chain analytics.
Snapp believes executives often aren’t comfortable with demand forecasting software because they don’t know how it works. Vendors could be to blame because they may not have explained the system well, the article says, whereas wholesale and retail distribution companies also may not have had a qualified person implementing their demand forecasting initiatives. But whoever is responsible for implementation must know how to use the technology and troubleshoot it and be able to “socialize the solution,” Snapp says.
One other major reason for lack of user adoption is the need for the users to set the parameters for the software. For example, some applications use projected demand of a product as a factor for calculation of required inventory replenishment; however, there’s an expectation that the user knows parameters such as how many months of past history are an accurate predictor of future demand. This number could also vary by item.
Asking users to do that kind of analysis is daunting, which is why the adoption of these kinds of systems can fail. There are so many parameters that must be fed into the system. In many cases, users don’t understand the parameters and what they mean, or they don’t understand their data well enough to be able to provide the values for those parameters.
Some newer applications do the work of analyzing the base data for the user to determine what inventory replenishment algorithms actually fit their data best. Most companies will not successfully implement this class of application unless the software will do all of the work by analyzing the data and setting the assumptions. And, of course, companies will also need proof that it works.
Source: DC Velocity, June 2013